DE102012221251A1 - Semantic and contextual search of knowledge stores - Google Patents

Abstract

A method, apparatus, and article for semantic and contextual searching of a knowledge store. The method includes generating a search query for each of the at least one term related to the target term to form a search context, the search query for each of the at least one related term having at least one word derived from at least one record of that term was previously created in the project, performing the search query on a search index of a knowledge store to recognize a record of the related term in which the search query was generated, and retrieving the at least one record of the target term from the memory as a search result in that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term returned as a result of performing the search query on the at least one record of the at least one related term.

Description

Field of the invention

Embodiments of the invention generally relate to information technology, and more particularly to searching knowledge stores.

Background of the invention

Service companies create knowledge for every job they take on. Success can often be driven by the quality of knowledge that is created during the assignment. In addition, success and failure are directly related to the relevant experience of the knowledge workers involved.

Data (or knowledge) is recorded in the different documents that are created during a job (contracts, specifications, design artifacts, code, etc.). Companies archive these documents as resources so that resources can be leveraged in future projects. The custody and reuse of this knowledge can translate individual knowledge into a business functionality. However, the efficient reuse of data may depend on the structure of the data in the memories as well as the simplicity and accuracy with which data is searched for and retrieved.

As requests are gathered, most data is created as unstructured documents such as Word documents, spreadsheets, and so on. Advisory activities often use document templates with a rudimentary structure; however, these are treated as unstructured text in resource storage.

Existing approaches include domain-specific environments that can help standardize content creation, preserve semantics and traceability between different artifacts, and store material as semi-structured metadata in Saving. Existing approaches also include the free text search using keywords in unstructured documents. However, such search results can be very inaccurate and return a large amount of irrelevant data.

Furthermore, existing enterprise search techniques are not nearly as effective as Internet search techniques. For example, link analysis techniques that are useful in getting high quality web pages are not helpful because of the lack of hyperlink structure in the in-house search.

Therefore, there is a need for efficient enterprise data store scanning techniques.

Brief description of the invention

In one aspect of the present invention, techniques for semantic and contextual searching of knowledge stores are provided. An exemplary computer-implemented method of semantically and contextually searching a knowledge store to provide a record of a target term based on a search term determined by records of at least one related term previously created in a project may be described in US Pat another step of generating a search query for each of the at least one term related to the target term to form a search context, the search query comprising at least one word for each of the at least one related term comprising at least one record of that term previously created in the project deriving the search query on a search index of a knowledge store to identify at least one record of the at least one related term for which the search query was created and retrieving the at least one record of the destination term from the memory as a search result, such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term returned as a result of performing the search query on at least one record of the at least one related term.

Another aspect of the invention, or elements thereof, may be implemented in the form of a product that materially embodies computer readable instructions that, when implemented, cause a computer to perform a variety of method steps as described herein. Further, another aspect of the invention, or elements thereof, may be implemented in the form of a device including a main memory and at least one processor connected to the main memory and operable to perform listed method steps. Yet another aspect of the invention, or elements thereof, may be embodied as means for carrying out the method steps or elements thereof described herein; The means may include, but are not limited to (i) a hardware module (s), (ii) a software module (s), or (iii) a combination of hardware and software modules Software modules; Any of (i) to (iii) implement certain techniques described hereinafter and the software modules are stored in a tangible computer-readable storage medium (or media).

These and other objects, features and advantages of the present invention will become more apparent from the following detailed description of illustrative embodiments of the description to be read in conjunction with the accompanying drawings.

Brief description of the drawings

1 FIG. 12 is a diagram illustrating exemplary semantic search retrievals and relevant results according to an embodiment of the present invention; FIG.

2 FIG. 12 is a diagram illustrating exemplary representations of recognized terminology according to an embodiment of the invention; FIG.

3 Fig. 12 is a diagram for illustrating an example of the search procedure according to an embodiment of the invention;

4 FIG. 10 is a block diagram illustrating an exemplary embodiment according to an aspect of the invention; FIG.

5 FIG. 10 is a flowchart illustrating techniques for semantic and contextual searching of knowledge stores in accordance with an embodiment of the invention; FIG.

6 FIG. 12 is a diagram illustrating the search algorithm according to one embodiment of the invention; FIG.

7 FIG. 10 is a flowchart illustrating techniques for semantic and contextual searching of a knowledge store according to an embodiment of the invention for providing a record to a goal term based on a search context established by records of at least one related term previously created in a project; and

8th FIG. 12 is a system diagram of an exemplary computer system on which at least one embodiment of the invention may be practiced.

Detailed description of embodiments

As described herein, one aspect of the present invention includes, among other things, semantic and contextual searching of knowledge repositories of service companies. Service companies create "knowledge" for every contract they take on. For example, the success of a job / project can be controlled by the quality of knowledge that is created during the job, and the success or failure can often be directly related to the relevant experience of the knowledge workers involved.

As used herein, the term "knowledge" may refer to data collected in the various documents produced (eg, contracts, specifications, design artifacts, code, etc.). Companies archive these documents as resources so that resources can be leveraged in future projects. The efficient custody and reuse of this knowledge can translate individual knowledge into business functionality. In addition, the use of this knowledge can serve as protection against employee turnover, new consultants can be updated more quickly, consultant productivity and the quality of solutions can be improved, etc. As described in detail here, the efficient and effective reuse of such depends However, data relies on the structure of the data in the stores as well as the simplicity and accuracy with which to search for and retrieve data.

One aspect of the invention includes, among other things, techniques for organizing and reusing knowledge resources, in particular artifacts associated with business processes, that have been developed in gathering requirements in a business transformation project. The knowledge resources are stored in a semi-structured format. In addition, it is known that much of the content of results of process-related work can be represented in a domain-specific data model. Thus, at least one embodiment of the invention includes, among other things, storing these work results in a semi-structured format (eg, Extensive Markup Language (XML)) based on a domain-specific schema.

For example, semi-structured data highlights different terms, such as process steps and business benefits, that exist in a process document. Therefore, recognizing keywords in the description of a particular term would make searching more efficient and accurate, especially in cases where process documents are not consistent in terms of expressions.

As also described herein, an aspect of the invention includes, among other things, a search strategy that uses the context of a given task to retrieve relevant data from memory. When searching for and reusing knowledge resources, keywords in the description of a particular term are recognized and translated into the "context" of a query. In contrast to just gathering a set of keywords as context, one aspect of the invention translates the context into rich semantic queries that take advantage of the semi-structured nature of the content.

When recognizing context, it should be noted that the tasks for creating data in consulting and service contracts are not independent of each other. The data created during any task may depend on the data that exists in one or more artifacts already created in the project. For example, in a typical application development job, the requirements are first worked out, then the business processes are graphed, and finally, the source code is developed. The source code depends heavily on the request documents and the process definitions. In a highly collaborative environment, knowledge workers often have no knowledge of the overall context behind their assigned task of creating data.

1 FIG. 12 is a diagram illustrating exemplary semantic search queries and relevant results according to one embodiment of the present invention. FIG. One of the key documents created during a project to transform a company is the process definition document (PDD). A PDD contains a detailed description of a business process and in particular information about different terms. These terms include, for example: (1) process: the name and a brief description of the process; (2) process steps: one or more steps that make up the process; (3) inputs: one or more inputs to the process steps; (4) Expenditure: one or more outputs from the process steps; (5) business benefits: one or more benefits to the business; (6) impact of Organizational Change Impact (OCI): impact of the process on the organization; (7) Requirements: some functional or non-functional requirements; (8) rules: rules and regulations that the process must comply with; (9) integration issues: various applicable technical or functional points in the integration; (10) important design decisions: considerations for the design of the process; (11) trigger: the trigger points of the process; and (12) dependencies: the dependencies of the process. Upon completion of a project, these documents are cleaned up to remove customer-specific data that has been placed in a knowledge store that is to be used to advantage in future projects.

To illustrate the search for data in such a memory, we consider the in 1 shown example in which three PDDs (in the lines) and three terms are mapped (in the column process name, process steps or OCI). All processes include the term "Purchase Orders" in their names; the process steps of PDD1 102 and PDD2 104 are very similar while the steps of PDD3 106 strongly different. In 1 is also the table 108 The search results derived from a simple keyword-based search are contrasted with a semantic search when searching with example keywords and queries (second column in table 108 ) on the three documents PDD1, PDD2 and PDD3.

Consider the scenario where a consultant needs to create the steps for an order process and understand and document the impact of the change on the process. To take advantage of the data collected in the knowledge store, the consultant searches the store for similar PDDs to verify the OCI data in those documents. If the search is performed using the keyword "Purchase Orders", all three PDDs (ie 102 . 104 and 106 ). Suppose the consultant mentions another key word, "Create Purchase". The search would still return all three documents. Essentially, the search lacks the ability to match keywords to terms, thereby searching a restricted range and not the entire PDD for a keyword (ie, a term).

Accordingly, one aspect of the present invention includes, among other things, extracting terms from PDDs and performing a search on terms. At the in 1 Among other things, an embodiment of the invention depicted could include the creation of two queries to search for: (1) "Purchase Orders" over the process (term) term and (2) "Create Purchase" over the process step term. When the queries are executed, only PDD1 will be executed 102 and PDD2 104 which are the PDDs the consultant wants to review. This improves the accuracy of enterprise search.

2 FIG. 12 is a diagram illustrating exemplary representations of recognized terminology according to one embodiment of the invention. FIG. In 2 are a concept 202 , a record 204 , Relationships 206 between terms and links 208 displayed between records. Furthermore, a data schema in the sense used here is an image of terms and their relationships with each other. For example, request, business process, steps may be terms in a schema, while (business process → step) may be a relationship. One aspect of the invention requires the presence of a domain data schema and memory that organizes the content according to such a scheme.

In addition, a record as used herein is an instance of a term that includes data for the term. It is assumed that a knowledge store is a collection of records of terms defined in the domain data schema. Data of a data set can be in the form of plain text, rich text or in binary form (images and attachments). In addition, two records created in the same project are considered related, R1 → R2, if their respective terms are related in the schema, ie C1 → C2. For example, in 2 The company process record "External production order being processed" is linked to the step record "Create purchase requisitions - external finished.".

Also, in the sense that the notion of similarity in records is used here, two sets of the same term are considered similar (R1 ~ R2) when there is a high degree of textual similarity. In text similarity, different scales such as cosine similarity may be used. From experience, an embodiment of the invention may assume that R1~R2 when the cosine similarity (R1, R2) is> k (where k is a certain threshold, for example 0.5).

As discussed, a consulting or service project can be thought of as a series of data authoring tasks, each task generating a record for a particular term. Accordingly, one aspect of the present invention, while creating a record for the term X, provides techniques for outputting relevant records from a knowledge store in which records from previous projects are stored.

The similarity in two records of a term can affect the similarity in the corresponding records of a dependent term. If, for example, a set of very similar records is found for requests, the corresponding records for processes should also be similar. Symbolically expressed, R _a ~ R _b leads to R _x ~ R _y when R _a → R _x and R _b → R _y . However, different terms may determine the content of a dependent term to varying degrees. For example, the desired benefit to the business may be more motivating than the requirements when choosing a process.

In order to effectively exploit the available context in the search, it is preferable to have an idea of how closely the similarity in records of one term determines the similarity in records of another dependent term. To quantify this relationship, one aspect of the invention involves calculating the following probability Rel on ordered term pairs (C _i , C _j ) as a probability that two records of C _j are similar, assuming their associated records of a related term C _i are known similarly. Expressed symbolically, Rel (C _i , C _j ) = P (R _x ~ R _y R R _a ~ R _b , R _a R R _x , R _b R R _y ), where R _a € C _i , R _b € C _i , R _x € C _j and R _y € C _j .

In one aspect of the invention, the knowledge organization may include. All records are indexed as documents in a full-text search engine, and the name of the term is included as a field in the document, so that semantic search is possible. In addition, scores of the pairwise dependency (denoted Rel) between all terms in the schema are calculated by measuring how the similarity in different records of one term correlates with the similarity in another term. These scores are used to rank search results.

Accordingly, in one aspect of the invention, the search procedure includes finding a search context, performing search queries, and ranking search results. When finding a search context for each term on which X (the new project) depends, a search query is generated by extracting the words from their records (if they have already been created). This set of queries forms the search context. For example, the generation of steps depends on the requirements and benefits for the business. Accordingly, queries are generated from the data sets of requirements and benefits for the enterprise to search for steps.

Also, any search query (as generated above) is performed on indexes (for example, using the full-text search) to obtain records of the To return the term used to generate the query. In addition, the records retrieved by X are linked to the query results in memory, and these records form the query results for X. For example, the lookups are run to obtain similar records of the requirements and benefits to the business. The step records associated with these records of requirements and benefits for the business are the search results.

In addition, a full-text search returns the search results to a relevance score on all query results returned by the full-text search (X-related records). One embodiment of the invention involves obtaining a rating to the records of X by weighting the relevance score to the query result using the Rel score between X and the related term from which the query was generated. In the sense used here, the Rel score is the extent to which related terms exert an influence on X. For example, by using Rel (business advantage, step) or Rel (requirement, step) weights the records of steps that are related to enterprise query results and requirements.

3 Fig. 12 is a diagram illustrating an example of the search procedure according to an embodiment of the invention. In 3 are a knowledge store 302 and an authoring environment 304 pictured. Furthermore, predefined data records for requirements (R) and advantages for companies (B) are represented by creation steps (P). Accordingly, one embodiment of the invention would involve searching for similar requirements and benefits for in-store businesses. In the example in 3 this would include R1 ~ R, R2 ~ R, B2 ~ B. The techniques would also include finding steps (P1 and P2) associated with R1, R2 and B2. The relevance rating of P2 can be calculated as the weighted average of Sim (B, B2) and Sim (R, R2), where the weights are Rel (Business Advantage, Step) and Rel (Requirement, Step). As mentioned above, text similarity (Sim) between data sets can be measured in a different standardized manner (for example, by cosine similarity or editing distances).

4 FIG. 12 is a block diagram illustrating an exemplary embodiment according to an aspect of the invention. FIG. In 4 the architecture of a search system of one aspect of the invention is depicted. The system is functionally divided into two parts, an indexing component and a search component. In at least one embodiment of the invention, the components may be mounted on a conventional Full Text Search (FTS) engine (eg, Apache Lucene) installed in a knowledge store.

With regard to the indexing procedure, as described herein, a typical work product in process advisory orders collects data across multiple terms. Rather than adding full work products as documents in a full-text search engine, one embodiment of the invention builds an index of records 424 on. For example, a PDD work product may be divided into lists of records of requirements, process steps, benefits to the business, etc., and each individual record may be indexed as a document. The one in a structured authoring environment 402 generated content is already in the form of the records 424 before and can therefore be indexed easily.

When a memory is a boot with a skeleton of existing documents 404 An aspect of the invention may be a data extraction module, which must be performed in formats such as MS Word 406 to break documents into records and then load them. Bestandsdokumente 404 be through the data extraction module 406 to transform it into a semantic format. Each document created from a record r _i lists the names of the term C _i as fields and links them to other documents of records n _{j such} that r _i → r _j . Documents are created using format-specific parsers, such as the parser module 408 analyzed, the origin of their content is determined, and the indexing module 410 becomes an FTS index 422 built up. Text in records is indexed through a full-text search engine. In addition, as part of the overall indexing procedure, one aspect of the present invention involves the Rel calculation module 412 calculates and stores values of rel for each pair of related terms in the schema. For example, the pairwise dependency between terms can be calculated.

With respect to the search procedure, the process of searching for relevant records for a term C _i by exploiting (and / or tracking) the context described in records of related terms proceeds as follows. The set of terms in each record of each related term of C _i is extracted to use the query generation module 414 to form a search query. This query is done using the search engine 416 via the FTS Index 422 and returns similar records of the same term the query was generated. In addition, records of the target term C _{i associated} with the search results are resolved and using the target fetch module 418 accessed. Further, these records are generated by C _i using the rank summary module 420 as final search results 426 returned.

• a) des Ausmaßes, in dem sich die Datensätze der mit dem Suchergebnis verknüpften verwandten Begriffe von C_i textlich den Datensätzen ähneln, die im Projekt bereits erstellt wurden; und
• b) des Ausmaßes, in dem die verwandten Begriffe von C_i, deren Datensätze zur Formulierung der Suchabfrage verwendet wurden, die Erzeugung eines Datensatzes von C_i beeinflussen können.

The ranking represents a relevance score of the search results based on the current content specified in the project. An embodiment (discussed below) calculates the relevance score as a function:

• a) the extent to which the records of related terms associated with the search result of C _{i are} textually similar to the records already created in the project; and
• b) the extent to which the related terms of C _i whose records were used to formulate the search query may affect the generation of a record of C _i .

5 FIG. 13 is a flowchart illustrating techniques for semantic and contextual searching of knowledge stores in accordance with an embodiment of the invention. FIG. As in 5 shown, two sub-areas of the techniques can be seen: knowledge organization and search procedure. With regard to knowledge organization, the step includes 502 accessing a data schema and a store of records (in a structured format such as XML) that match the input schema. The step 504 involves indexing the records as documents in a full-text engine and adding the semantics (term labels) as searchable fields in the document. The indexed records will be in step 510 inverted full-text indexes.

The step 506 involves calculating the text similarity "Sim" to each term between all record pairs (R1, R2) of that term in the data store. For (R1, R2)> k, R1 is R2. The value of the threshold k can be adjusted experimentally. The step 508 involves calculating Rel (C _i, C j) = P (R _x ~ R _y R R _a ~ R _b , R _a R R _x , R _b R R _y ) for each pair of terms Ci and Cj, where R _a € C _i , R _b € C _i , R _x € C _j and R _y € C _j . In step 512 For all pairs of terms a rel-score is provided.

With regard to the search procedure, the step includes 514 Obtaining a draft term X as well as context, ie records of terms related to X. The step 516 involves generating a query for each record in the context, with the words being extracted from the record. The step 518 involves performing a full text search for queries in Q to get a set of query results for each query along with their query's similarity score. The search is used to retrieve records of the same term as in the query. A query result can be represented as a triple: q - r = <data set, term, score>. The step 520 Also, for each query result q-r, it includes querying any record Rx of X that may be associated with its record in memory, and generating a target query result as <Rx, q-r. term, q-r.point> which is added to the set of goals T.

The step 522 For each record R that exists in a particular destination, summarizing the rank of the record includes the following: Initialize, SumRang = 0, SumGweight = 0 for each target is t € T such that t .Date = R
begin
SumRang = SumRang + t.Point Number × Rel (t.Concept, X)
SumWeight = SumWeight + Rel (t.Concept, X)
end
Rank = SumRang / SumGeweight

The target record is added to the set of search results along with its rank. Further, the step includes 524 spending a lot of search results {<record of X, rank>}.

6 is a schematic illustrating the search algorithm 602 according to an embodiment of the invention. In a query to retrieve records of the term C _i , "search process" considers the set of available records of related terms as search context βC _i and accepts a set of user-defined keywords q _u as an optional input.

"Query formulate" (lines 17-23) generates a set of queries Q from βC _i . A query is represented as a 2-tuple (r ', C), where r' is a set of expressions and C is the term whose records are searched should be. For a query derived from r _i ε βC _i , qr 'contains the expressions in r _i .δ, and qC is r _i .C. To the user-specified keywords q _u , a query is generated with the expressions in q _u (line 22), and q.0 is assumed to correspond to c _i . Next, each query q ε Q is triggered on the full-text search index (FTS index) (line 5) generated for all records in memory. S _i denotes the set of first search results for q; it lists the data sets of the term qC together with their relevance score numbers α, so that α is greater than a threshold value κ (line 15).

In addition, one aspect of the invention includes the notion of a goal. An object contains a record t r _i of C _i, that is associated with a data record r, which listed in a search result s ε Si, that is, r _i r →. The relevance score α of the search result s and the term rC (line 24) are also stored in t. Further, in one embodiment, the rank summarization aspect of the invention "rank summarize" (lines 26 to 30) recalculates the rank of the target records captured in all search results of all queries q ε Q. The relevance score of a record {t _i } present in the set of targets is adjusted as a weighted average of the original relevance score t _i .α, where the weights are values of Rel (t _i .C, C _i ). To return the end results, the set of target records is sorted in descending order.

7 5 is a flowchart illustrating techniques for semantic and contextual searching of a knowledge store according to an embodiment of the present invention to obtain a (relevant) record on a search context established by records of at least one related term previously created in a project to provide the (desired) target concept. The step 702 involves generating a search query for each of the at least one term related to the target term to form a search context, the search query for each of the at least one related term having at least one word derived from at least one record of that term, that was previously created in the project. This step may be performed, for example, using a query generation module. Generating a search query for each of the at least one term on which a project relies to form a search context involves generating a set of at least one query from text of each record created in the project that affects a current activity of the project.

The step 704 involves executing the search query on a knowledge store search index (eg, via a text search engine) to identify at least one record of the at least one related term at which the search query was generated. This step may be performed, for example, using a search engine. The knowledge store contains records from at least one previous project, and the records from the previous project or projects are defined in a data schema. A relationship between two or more terms in the records is also defined in the data schema.

In one aspect of the invention, executing the search query on the search index of a knowledge store to recognize in memory at least one record associated with at least one term on which the project is based may include analyzing data generated in the project that is associated with a current activity, and the data is used to recognize the contents in the memory.

The step 706 includes retrieving the at least one record of the target term from the memory as a search result such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term that results from performing the search query on the at least one a record of the at least one related term is returned. This step may be performed using, for example, a destination polling module. Retrieving the at least one record from memory as a search result to the project may include retrieving a record associated with the at least one recognized record in the memory for a related term and similar to the record of the related term previously created in the project , In addition, retrieving the at least one record from memory as a search result to the project may include temporarily tracking links originating from recognized records of the related term in memory that resemble the at least one record of the related term previously created in the project was used to retrieve at least one record of the target term.

In the 7 The techniques depicted also include generating a rank for the at least one retrieved record of the target term by weighting a relevancy rank returned from the search index search to the at least one related-item record using a Rel score between the target term and the at least one a related term, where the Rel score is a measure in which the related term record previously created in the project can affect the generation of a record to the target term in the project. The rank uses a pre-computed dependency score Rel between terms in a data schema in the knowledge store. In addition, generating a rank to the at least one retrieved record includes calculating a rank for each retrieved record as a function of a score indicating the textual relevance of each record in the search query that resulted in the search result. Furthermore, generating a rank for the at least one record includes calculating a rank for each retrieved record as a function of a score indicative of the extent to which a goal term depends on each of the at least one term upon which a project used to formulate the search query is attributed led to the search result.

As also described herein, generating a rank may include computing a pairwise dependency score between all pairs of terms (ie, all terms for which records are available in memory) by analyzing their records in the knowledge store. Computing a pairwise dependency between terms involves, for each term, computing a text similarity score for all pairs of records for each term in memory. A record pair is determined to be similar if the text similarity score for a record pair is greater than a selected threshold. Moreover, calculating a pairwise dependency score between terms may involve calculating a probability for all pairs of terms (C _i , C _j ) that two sets of C _j are similar, provided that records of a related term C _i are known as similar. The probability can serve as a score for the dependency with which C _j depends on C _i .

In the 7 The techniques illustrated may also include, as described herein, providing a system wherein the system includes its own software modules and each of its own software modules is embodied on a tangible computer readable recordable storage medium. All modules (or any subset thereof) may reside on the same medium, or each module or subset may reside on a different medium, for example. The modules may include any or all of the components shown in the figures. In one aspect of the invention, the modules include a rel-calculation module, a query generation module, a target fetch module, and a rank summary module, which may be executed on a hardware processor, for example. The method steps may then be performed as described above using the system's own software modules running on a hardware processor. Further, a computer program product may include a tangible computer readable writable storage medium having code that is configured to execute at least one method step described herein, including providing the system with its own software modules.

In addition, the in 7 can be implemented using a computer program product, which may include computer usable program code stored on a computer readable storage medium in a data processing system, and wherein the computer usable program code has been downloaded via a network from a remote data processing system. Additionally, in one aspect of the invention, the computer program product may include computer-usable program code stored on a computer-readable storage medium in a server computing system, and wherein the computer-usable program code is downloaded over a network to a remote computing system for accessing the remote system in FIG a computer-readable storage medium to be used.

Those skilled in the art will appreciate that aspects of the present invention may be embodied in the form of a system, method, or computer program product. Accordingly, aspects of the present invention may take the form of a fully hardware embodiment, an entirely software embodiment (including firmware, resident software, microcode, etc.) or an embodiment that combines software and hardware aspects generally referred to herein "Circuit", "module" or "system". Further, aspects of the present invention may take the form of a computer program product embodied in a computer readable medium having computer readable program code embodied thereon.

An aspect of the invention may be implemented in the form of a device including a main memory and at least one processor connected to the main memory and operable to perform exemplary method steps.

In addition, one aspect of the present invention may utilize software executing on a general purpose computer or on a workstation. With reference to 8th For example, such a realization may be a processor 802 , a main memory 804 and use an input / output interface, for example, by a display unit 806 and a keyboard 808 is formed. The term "processor" as used herein should be understood to include all processing units, e.g. B. a processing unit with a CPU (CPU = Central Processing Unit) and / or other forms of processing circuits includes. Furthermore, the term "processor" may include one or more individual Designate processors. The term "memory" includes all memory types associated with a processor or CPU, for example random access memory (RAM), read only memory (ROM), fixed memory (e.g. a hard disk drive), a removable storage unit (for example, a floppy disk), flash memories, and the like. In addition, the term "input / output interface" as used herein should be understood to include, for example, a mechanism for inputting data to the processing unit (for example, a mouse) and a mechanism for providing results related to the processing unit (e.g. Example, a printer). The processor 802 , the main memory 804 and the input / output interface such as the display unit 806 and the keyboard 808 can communicate with each other, for example, over the bus 810 as part of the data processing unit 812 be connected. Suitable compounds, for example via the bus 810 , also can use a network interface 814 such as a network card, which may be provided as an interface to a computer network, as well as a media interface 816 such as a floppy disk or a CD-ROM drive that interfaces with that in the medium 818 can be provided.

Accordingly, computer software with instructions or code for carrying out the methodologies of the invention as described herein may be stored on associated storage devices (eg, ROM, fixed or removable storage) and partially or fully loaded by a CPU (eg, in RAM) in the ready state. and be implemented. Such software could include, but is not limited to, firmware, resident software, microcode and the like.

A data processing system suitable for storing and / or executing program code includes at least one processor 802 that via a system bus 810 directly or indirectly with memory elements 804 connected is. The memory elements may include local memory used during the actual implementation of the program code, mass storage, and cache memory that provides a temporary storage capability for at least a portion of the program code to reduce the number of code calls from the mass memory during the conversion.

Input / output units or I / O units (including, but not limited to, keyboards 808 , Display units 806 , Pointing devices and the like) can be either directly (such as via the bus 810 ) or through intermediate I / O controllers (which have been omitted for clarity).

Network adapters such as the network interface 814 can also be connected to the system so that the data processing system can be connected via intermediate private or public networks to other data processing systems, remote printers or storage units. Modems, cable modems, and Ethernet cards are just some of the types of network adapters currently available.

As used herein, including the claims, a "server" is a physical data processing system (e.g. 8th shown system 812 ) on which a server program is running. It is understood that such a physical server may include a display unit and a keyboard.

As mentioned, aspects of the present invention may take the form of a computer program product embodied in a computer readable medium having computer readable program code embodied thereon. Also, any combination of one or more computer-readable media may be used. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. Computer-readable storage media may include, but are not limited to, an electronic, magnetic, optical, or electromagnetic system, or an infrared or semiconductor system or device or device, or any suitable combination of the foregoing. More specific examples (incomplete list) of computer-readable storage media include, but are not limited to, an electrical connection to one or more lines, a portable computer disk, a hard disk, Random Access Memory (RAM), Read Only Memory (ROM), an erasable programmable Read-only memory (EPROM or flash memory), an optical fiber, a compact disc (CD-ROM) read only memory, an optical storage unit, a magnetic storage unit, or any suitable combination of the foregoing. In the context of the present document, a computer readable storage medium may be any objectified medium that may contain or store a program that may be used by or in connection with a command execution system, apparatus, or device.

A computer readable signal medium may include, among other things, a data signal transmitted in baseband or as part of a carrier wave having computer readable program code embodied therein. Such a transmitted signal may take any of a variety of forms including, but not limited to, an electromagnetic or optical form, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that may transmit, transmit, or transport a program that may be used by or in connection with an instruction execution system, apparatus, or device ,

Program code embodied on a computer-readable medium may be transmitted using a suitable medium, including, but not limited to, wireless, wired media, optical fibers, RF, etc., or using any suitable combination of the foregoing.

Computer program code for performing operations in aspects of the present invention may be written in any combination of at least one programming language, including in an object-oriented programming language such as Java, Smalltalk, C ++ or the like, and in conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a Local Area Network (LAN) or Wide Area Network (WAN), or the connection may be made to an external computer (e.g. over the Internet using an internet service provider).

In the present document, aspects of the present invention are described with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowcharts and / or block diagrams and combinations of blocks in the flowcharts and / or block diagrams may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing device to generate a machine such that the instructions executed via the processor of the computer or other programmable data processing device include means for realizing in a block Create functions / actions specified for the blocks of the flowchart and / or the block diagram.

These computer program instructions may also be stored in a computer readable medium that may instruct a computer, other programmable computing devices, or other entities to function in a particular manner such that the instructions stored in the computer readable medium create a product having the instructions corresponding to those in the instructions implement a block or function / action specified in the blocks of the flowchart and / or block diagram. Therefore, one aspect of the invention includes a product that materially embodies computer readable instructions that, when implemented, cause a computer to perform a variety of method steps as described herein.

The computer program instructions may also be loaded into a computer, other programmable computing devices, or other devices to cause a number of operations to be performed on the computer, other programmable devices, or other devices to provide a computer-implemented process so that the instructions executed on the computer or on other programmable devices provide processes for realizing the functions / actions specified in a block or blocks of the flowchart and / or block diagram.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Accordingly, each individual block in the flowchart or in the block diagrams represent a module, a component, a segment or a part of the code which has at least one executable instruction for realizing the specified logic function or logic functions. It should also be noted that in some alternative implementations, the functions specified in the block may be performed in an order other than that indicated in the figures. For example, two consecutive blocks may in fact be executed substantially concurrently, or the blocks may sometimes be executed in reverse order depending on the functionality involved. In addition, it should be noted that each block of the illustrated block diagrams and / or the illustrated flowchart as well as combinations of blocks in the illustrated block diagrams and / or in the illustrated flowchart using special systems based on hardware to perform the specified functions or actions combinations of special hardware and computer instructions can be realized.

It should be noted that any of the methods described herein may include an additional step of providing a system having proprietary software modules embodied on a computer-readable storage medium; For example, the modules can be any or all of the 4 components shown included. The steps may then be as described above using the on a hardware processor 802 executed own software modules or submodules of the system. Further, a computer program product may include a computer-readable storage medium having code that is configured to be implemented to perform at least one method step described herein, including providing the system with its own software modules.

In any event, it will be understood that the components illustrated herein may be implemented in various forms of hardware, software or combinations thereof, for example in the form of application specific integrated circuits (ASICS), function specific circuits, in the form of a suitably programmed general purpose digital computer Memory and the like. In view of the teachings of the invention set forth herein, those skilled in the art will be able to envision other forms of implementation of the components of the invention.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms "a / a" and "the" used herein also include plurals, unless expressly stated otherwise in the context. It should also be understood that the terms "having" or "having" as used in this specification refer to the presence of specified features, integers, steps, operations, elements, and / or components, but the presence or addition of others Do not exclude features, integers, steps, operations, operations, elements, components, and / or groups thereof.

The corresponding structures, materials, actions, and equivalents of all means (step plus function elements) in the following claims are intended to include all structures, materials, or acts for performing the function in conjunction with other elements expressly claimed , The description of the present invention is intended to be illustrative and descriptive, but is not exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the invention. The embodiment has been chosen and described in order to best explain the principles and practice of the invention and to enable others skilled in the art to appreciate various embodiments of the invention having various modifications suitable for the particular use contemplated.

At least one aspect of the present invention can provide a useful effect, such as ranking search results, by weighting the textual relevance of the content based on the strength of the relationship between the semantics of the executed query and the content being created.

The descriptions of the various embodiments of the present invention are intended to be illustrative, but not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those skilled in the art without any departure from the scope and spirit of the described embodiments. The terminology used herein has been chosen to best explain the principles of the embodiments, the practical application or technical improvement over the technologies found on the market, and to enable others with appropriate skill in the art to understand the embodiment disclosed herein.

Claims (15)

A method for searching a knowledge store semantically and contextually to find a knowledge store based on a search context defined by records of at least one related term previously created in a project To provide a record to a target term, the method comprising: generating a search query for each of the at least one term related to the target term to form a search context, the search query having at least one word for each of the at least one related term derived from at least one record of that term previously created in the project; Executing the search query on a knowledge store search index to identify at least one record of the at least one related term to which the search query was generated; and retrieving the at least one record of the target term from the memory as a search result such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term resulting from executing the search query on the at least one Record of the at least one related term is returned; wherein at least one of the steps is performed by a computer unit. The method of claim 1, further comprising: Generating a rank for the at least one retrieved record of the target term by weighting a relevancy rank returned from the search on the search index to the at least one related-term record using a Rel score between the target term and the at least one related term, where the Rel score is a measure in which the related term record previously created in the project can affect the generation of a record to the goal term in the project. Method according to one of the preceding claims 1 to 2, wherein the rank uses a precalculated score of Rel dependency between terms in a data schema in the knowledge store and / or generating a rank to the at least one retrieved record computing a rank for each retrieved record as a function of a score indicating the text relevance of each record in the search query that resulted in the search result, and / or generating a rank to the at least one record as calculating a rank for each retrieved record a score indicative of the extent to which a target term depends on each of the at least one term upon which a project is used to formulate the search query that resulted in the search result. The method of claim 2, further comprising: Compute a score of a pairwise dependency between all pairs of terms by analyzing their records in knowledge store. The method of claim 4, wherein calculating a pairwise dependency between terms comprises calculating a text similarity score to each term for each pair of records for each term in memory. The method of claim 5, further comprising: Identifying a record pair as similar if the text similarity score for a record pair is greater than a selected threshold. The method of claim 4, wherein calculating a pairwise dependency score between terms comprises calculating, for all term pairs (C _i , C _j ), a probability that two sets of C _j are similar, provided that records of one related term C _{i are known} as being similar. The method of claim 7, wherein the probability is a score for the dependence with which C _j depends on C _i . The method of claim 1, wherein the knowledge repository comprises records from at least one previous project. The method of claim 9, wherein the records from the at least one previous project are defined in a data schema. The method of claim 10, wherein a relationship between two or more terms in the records is defined in the data schema. The method of claim 1, wherein executing the search query on the search index of a knowledge store to recognize in memory at least one record belonging to at least one term on which the project is based comprises analyzing data generated in the project, the lead to a current activity, and the data is used to recognize the content in the memory and / or generating a search query for each of the at least one term on which a project is based to form a search context has a set of at least one query of text from each record created in the project affecting a current activity of the project, and / or retrieving the at least one record from memory as a search result to the project retrieving a record comprising the at least one recognized record is linked to a related term in the memory and the Resembles a record of the related term previously created in the project and / or retrieving the at least one record from memory as a search result to the project, temporarily tracking links originating from recognized records of the related term in memory, that are similar to the one or more records of the related term previously created in the project to retrieve at least one record of the target term. A system for semantically and contextually searching a knowledge store to provide a record to a target term based on a search context set by records of at least one related term previously created in a project, comprising: own software modules, each own software module embodied on a tangible computer readable medium; a main memory; and At least one processor connected to main memory and ready to execute steps using its own software modules, including: generating a search query for each of the at least one term related to the target term to form a search context, the search query for each of the at least one related term having at least one word derived from at least one record of that term; previously created in the project, and wherein generating a search query is performed by a query generation module executing on the at least one processor; executing the search query on a search index of a knowledge store to identify at least one record of the at least one related term to which the search query was generated, wherein the search query is performed by a search engine executing on the at least one processor; and retrieving the at least one record of the target term from the memory as a search result such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term that results from performing the search query on the at least one A record of the at least one related term is returned, wherein retrieving the at least one record is performed by a target fetch module that is executed on the at least one processor. A product comprising a computer readable storage medium having materially embodied computer readable instructions that, when implemented, cause a computer to perform a plurality of method steps, comprising: Generating a search query for each of the at least one term related to the target term to form a search context, the search query for each of the at least one related term having at least one word derived from at least one record of that term previously was created in the project; Executing the search query on a knowledge store search index to identify at least one record of the at least one related term to which the search query was generated; and Retrieving the at least one record of the target term from the memory as a search result such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term resulting from performing the search query on the at least one record of the at least one related term. A system for semantically and contextually searching a knowledge store for providing a record to a target term based on a search context set by records of at least one related term previously created in a project, comprising: at least one proprietary software module, each of its own software modules embodied on a tangible computer-readable medium; a main memory; and At least one processor connected to the memory and ready for operation: Generating a search query for each of the at least one term related to the target term to form a search context, the search query for each of the at least one related term having at least one word derived from at least one record of that term previously was created in the project; Executing the search query on a knowledge store search index to identify at least one record of the at least one related term to which the search query was generated; and Retrieving the at least one record of the target term from the memory as a search result such that the at least one retrieved record of the target term in the knowledge store is associated with a record of the at least one related term resulting from performing the search query on the at least one record of the at least one related term.

Images